Reliable monitoring of CO₂ storage reservoirs requires a combination of different observation methods. However, history matching is typically limited to CO₂ pressure data alone. This paper presents a multi-physical inversion of hydraulic pressure, CO₂ pressure, CO₂ arrival time and geoelectrical crosshole observations of the Ketzin pilot site for CO₂ storage, Germany. Multi-physical inversion has rarely been reported for CO₂ storage reservoirs. In contrast to previous studies, there is no need for pre-inversion of geophysical datasets as these are now directly included in a fully coupled manner. The deteriorating impact of structural noise is effectively mitigated by preconditioning of the observation data. A double regularisation scheme provides stability for insensitive parameters and reduces the number of required model runs during inversion. The model shows fast and stable convergence and the results provide a good fit to the multi-physical observation dataset. It has certain predictive power as the known migration direction of the CO₂ plume is captured. These results clarify two long discussed issues of the Ketzin CO₂ storage reservoir: (1) The pre-existing hypothesis of an existing hydraulic barrier became unsubstantial as the data series suggesting weak hydraulic communication are identified as erroneous. (2) Salt precipitation around the injection well doubles the injection overpressure compared to salt free conditions, which is equivalent to a well skin of 10. The presented framework allows to integrate various types of observations into a single multi-physical model leading to an increased confidence in the spatial permeability distribution and, in perspective, to improved predictive assessments of CO₂ storage reservoirs.

要可靠地监视CO ₂储存库，需要结合使用不同的观察方法。但是，历史记录匹配通常仅限于CO ₂压力数据。本文介绍了德国Ketzin储CO ₂试点的液压，CO ₂压力，CO ₂到达时间和地电横孔观测值的多物理场反演。很少有关于CO _2的多物理倒置的报道储水库。与以前的研究相比，地球物理数据集无需进行预反演，因为这些数据现在已以完全耦合的方式直接包含在内。通过对观测数据进行预处理，可以有效地缓解结构噪声不断恶化的影响。双重正则化方案为不敏感的参数提供了稳定性，并减少了反演期间所需模型运行的次数。该模型显示了快速稳定的收敛性，结果为多物理观测数据集提供了很好的拟合。由于捕获了CO ₂羽流的已知迁移方向，因此它具有一定的预测能力。这些结果澄清了Ketzin CO _2的两个长期讨论的问题储油库：（1）现有的水力屏障的假设变得微不足道，因为表明弱水力连通性的数据系列被确定为错误的。（2）与无盐条件相比，注入井周围的盐沉淀使注入超压成倍增加，相当于10口井的表皮。提出的框架允许将各种类型的观测值整合到单个多物理模型中，从而增加对空间渗透率分布的信心，以及从角度上看，可以改善对CO ₂储层的预测性评估。